Software testing company SmartBear Software Inc. is pushing its BearQ agentic quality assurance system beyond test automation with two new capabilities released today.
Both additions target the part of quality assurance that artificial intelligence has made more difficult by speeding up coding. SmartBear calls them Sources and Workflows.
Automating individual tests is no longer the bottleneck, in the company’s telling. The tougher job now is working out what still needs testing across sprawling systems, much of it already covered by tests sitting in other tools.
Sources widens BearQ’s view past the application itself. Requirements and bug reports now flow in from Atlassian Corp.’s Jira so the agent knows what ought to be tested. SmartBear’s own Reflect and Zephyr testing tools show what existing tests already cover. BearQ draws on that context to find gaps in coverage before it writes anything new, so teams end up with fewer duplicate tests.
Workflows picks up from there, letting a team chain jobs together behind a single trigger. Once a build lands in the continuous integration and delivery pipeline, for example, BearQ can spin up a test environment and run regression tests against it. Whatever the run turns up goes back into the team’s connected systems, and anyone who would rather not tie testing to builds can put the same sequence on a schedule. How much of a workflow runs without human review is up to the team.
Progressive autonomy is SmartBear’s name for that arrangement. Teams begin with people signing off on what BearQ does, then hand over more of the process as their confidence grows. Trust was the top barrier to adopting or scaling autonomous testing, named by 23% of the 1,436 U.S. and U.K. respondents in a survey SmartBear released Sept. 30.
Bridges Smith, vice president of product management for AI at SmartBear, called autonomous testing “the first step.” By Smith’s account, the bigger opportunity is connecting how an application actually behaves with what a team intends it to do and the test suite it has built up. People would still set the guardrails, Smith said, describing the shift as a move “beyond autonomous testing toward autonomous quality.”
SmartBear has been building BearQ out since it first appeared in March, when the company opened early access for web applications built by midmarket software-as-a-service providers. A September update made it an assignable agent inside Jira. SmartBear’s older lineup includes the Swagger tools for designing application programming interfaces, and about 32,000 organizations use its software, by the company’s count.
More Sources connections are due before the quarter ends. Both features get a public demonstration Thursday in a livestream with “Project Hail Mary” author Andy Weir.